OddThinking

A Map

Let me take you on a scenic walk through the world of business presentations. Even though we are only going to visit a tiny portion of this world, it is going to be a long hike. Let me give you a little map of where we are going, so you don’t get lost.

I am going to start by describing the ideal scenario – the mythical Garden of Eden of this world.

Then I am going to reveal to you three of the key sources of pollution into this world. I am going to show some of the far-reaching consequences of these pollutants – how they negatively impact communications and analysis.

I aim to convince you that these basic problems are the root cause of many business communication failures.

Then I will introduce the arguments of Edward R. Tufte. He will point to a snake in this garden, and denounce it as the devil. He will name the snake PowerPoint, and claim it causes all of the evil in this world.

I will walk you through Tufte’s arguments, and suggest that the snake isn’t the devil, just a natural part of the environment. While we should be careful not to be bitten by the snake, it is man-made pollutants that we should really work on.

I hope that when we return you will agree with me that we should focus our concerns away from PowerPoint, and towards the real causes of poor communication.

Garden of Eden

Like in real-world businesses, in our platonic Garden of Eden, decisions frequently need to be made based on technical or business analysis of an issue. Typically, a team of one or more experts is asked to produce an analysis, and report back to the management team who has the authority to make a decision.

The experts select between a wide-range of techniques to perform the analysis. Ultimately, all of the details of the analysis are produced in a detailed written report.

That report is then carefully peer-reviewed by other experts. The reviewers set out to find flaws in the analysis. They meet with the original authors to discuss the issues. They only let it past when they are happy to commit to its accuracy. They are aware of the dangers of “groupthink” and strive to avoid it.

If the report passes that gauntlet, a summary is produced. The summary explains the benefits and the costs, risks and recommendations. (The summary itself is peer-reviewed, too!) That summary is communicated to the managers through a combination of a presentation (with handouts), copies of the report and Question-and-Answer sessions. Follow-up sessions ensure that the information is understood and retained.

The presentations are, of course, filled with clear and accurate information. They are presented carefully to ensure that the managers had a good understanding of the issues. The experts have no difficulty with this, as they are trained, experienced and skilled presenters.

The managers may have access to additional facts outside the scope of the study. Great care is taken that those facts are not required for the original analysis. (For example, the management may have a greater understanding of business priorities, alternative proposals from other teams, etc.)

The managers use both the summary, and the additional facts, to make their wise and carefully considered decision.

Constraints of the Real World

In the real world, there are three “key drivers” or forces that prevent us from meeting the ideals described above.

Time is Limited

People are unable to spend as much time as they like on a project, and therefore must prioritise tasks to perform only the most productive ones. Of course, there is limited time available to perform that prioritisation, too.

People’s communication skills are typically poor.

Effective communication is difficult. Many people called upon to do presentations are unskilled, whether it is due to lack of training, lack of experience or lack of talent.

Overhead projectors have limited resolution

Compared to the resolution of paper, overhead projectors (including data projectors, televisions, rear-screen projectors, etc.) have very poor resolution. It is impossible to include the level of detail that is possible through other means, such as handouts.

Impact of the Constraints

I consider the three constraints mentioned above to be self-evident axioms. (Please let me know if you disagree.)

From these three simple axioms, it is possible to explain a number of different human behaviours:

Managers Deprecate Detail

A lesson that I learnt many years ago is that, as people are promoted upwards in management, there is a corresponding decline in the interest in “the details”. As people have a higher purview, they demand shorter and less detailed “Executive Summaries”.

This is a consequence of time limits. With more issues within a domain to be managed, there is less time to spend on each of them. It is necessary to delegate full analysis of a problem to underlings

A result of this trend is that managers focus on receiving their information through high-level summaries, and not through detailed reports.

In order to be successful in this strategy, managers must make two assumptions:

• that the presenters (i.e. experts) have performed the full analysis carefully, and
• that the devil is not in the details – that the summary accurately conveys the results of the analysis.

There are certainly times when each of these assumptions fails.

Managers exhibit short-attention spans

A secondary effect of the above need to avoid being caught up in details is that (at least when the above assumptions turn out to be true) managers are rewarded for having short attention spans – for not digging down into each and every issue and worrying away at it until the full story is understood, but accepting a sound-bite as sufficient understanding and moving on.

Is it negative to have a short attention span? It sounds bad. For many roles it certainly is bad. However, my point is that, in some roles where it is necessary to share your time amongst many simultaneous issues, it might well have some benefits too.

Slides given undue priority over other document formats

In order to meet time constraints, heuristics are used by busy managers to determine what should be given their attention. The PowerPoint slide format is correlated with quick “summary-level” information. The Word document format is correlated with longer, more detailed reports.

As a result, in my experience, managers will use the format of the document (e.g. .doc extension versus .ppt extension) as a heuristic to help assess if they have time to read a document.

Conversely, someone trying to attract the attention of management is under pressure to “game the system” by producing PowerPoint slides rather than Word documents, even for detailed documents when a presentation format is not the most appropriate one for the job.

Analysis Shortcuts Taken

If an audience of managers doesn’t read the full analysis (due to time constraint; see above), and an expert is also time-limited, there is a temptation to spend less time on the analysis and more time on the summary. This is a serious risk; if the measure of productivity is the quality of the slides, rather than the quality of the report, there can be a strong pressure to skimp on the analysis.

Bullet points instead of sentences

The low resolution of the projectors implies there is limited room on each page.

Bullet points are an effective means to cram more concepts per page, by discarding a large level of the natural redundancy found in human languages.

This has several downsides as well:

• Redundancy is used for error-checking; removing redundancy makes it more likely that the language will be ambiguous or misread.
• Removing the niceties (“syntactic sugar”) of language makes it less pleasant to read; it takes more effort to interpret.
• Statements about the real world often have a string of conditions attached – exceptions where the statement is not entirely true. In the crammed world of a bullet point, there is a temptation to drop these additional conditions, even if it results in over-simplification.

Use of bullet points instead of other structures

In a hand-drawn overhead projector slide, the cost of bullet points and the cost of boxes and arrows are about the same amount of effort.

Modern computer formats (including RTF, HTML and PowerPoint) have made bullet points remarkable cheap, especially in comparison with other structures.

Given such a price list, there is a strong impetus to use bullet points as a way of laying out a slide, even if it isn’t the ideal method. Such a compromise may result in less clear slides. Sometimes such a risk is unacceptable, but sometimes it is.

Minimal Copy-Editing

One place where limited time can often be saved is in the copy-editing.

Little time might be spent on spell-checking, ensuring consistency between terms, improving layout, etc.

The result may lower the clarity and perceived professionalism of the talk, but for a single-use presentation with a limited audience, the time may well be better spent on other tasks.

AutoContent, Design Templates, Corporate Branding Standards

One solution to a shortage of time to format data is to re-use design templates. These help to make the slides look consistent (especially for circumstances where branding is important), attractive and easy-to-read.

One solution to a shortage of time to analyse the data and a shortage of skills in constructing a persuasive argument, is to use AutoContent features and slide deck templates. These help the user to marshal their arguments into a cohesive whole.

These template tools have a benefit: they increase the chance that your colours, design and argument construction will be acceptably good. They also have a cost: they reduce the chance that your colours, design and argument construction will be exceptionally good.

Grabbing Attention Through Distractions

Depending on the circumstances, it can often be an effort to maintain the focus of the audience with short attention spans. I would lump a number of common techniques together as trying to “distract” the audience into paying attention – inserting various amusing bits of non-content to bring the audience’s focus back to the presenter and the real content.

The oft-repeated advice to start a speech with a joke suggests a way to bring the focus to the presenter. Pictures and animations, slide transitions, bright colours, variety in the fonts and page structure – these are all different distractions that can be judiciously used to spark up some dull content.

The downside is that if you use these techniques too much, or in inappropriate circumstances, the result is to distract away from your content, rather than towards your content.

They may be well-intentioned, but these techniques can reduce communication rather than enhance it.

Form Can Be Judged Faster Than Content

The quality of a slide deck is correlated to the amount of time you spend on the deck.

The use of figures, animations, colours takes time – that means that these items are correlated to the amount of time you spend on the deck.

So, if you don’t have time to review the content of a deck, you can use an overly-simplistic heuristic: the more figures, animations and colours that have been used, the better the quality of the deck.

I have watched as complete decks of technical slides that I had spent hours on producing have been strongly praised after sub-second evaluations. In return, I have learnt to game this system – ensuring that slides are colourful enough pass the “sub-second muster”, and giving the impression that I am a proficient slide producer. I often wrestle with myself over changes that will make the slide look fancier at the cost of clarity. I try not to cross too far over that line.

Meaningless Charts

Thinking through the best way to display some numeric data takes time and some communication skills.

When these two items are lacking, the resulting charts can be confusing, meaningless or even misleading.

PowerPoint Contribution to Evil

Looking at each of these issues, we can see that PowerPoint has only a peripheral impact on these evils.

PowerPoint provides AutoContent and similar templates to assist in guaranteeing adequate colour schemes.

PowerPoint makes bullet points much easier, without making other structures equally easy, increasing the tendency to use bullet points.

PowerPoint makes it easy to animate slides and transitions, insert pictures and to change fonts and colours. This increases the risk of inserting too many such devices, and distracting from the content.

In summary, by making some things very easy, PowerPoint increases the risk of over-using those features – hardly an unforgivable act!

A Rest Stop

So far, I have taken you for a walk through the world of business presentations, and hopefully convinced you that there is a simple model – three (or maybe four) assumptions – that explain many serious problems with business presentations.

In this model, the blame for poor quality presentations is slated to the poor workman (and woman) rather than the tool (i.e. PowerPoint).

Now, it is time to turn back, but this time, we will follow the footsteps of Edward R. Tufte.

So who is Tufte? Tufte is a former professor at Yale and a respected expert in presentation of statistical data.

He has published a number of books (that I haven’t read), and a number of essays (of which I have read a few).

In 2003, he caused a small stir by publishing an essay (available in booklet form) called The Cognitive Style of PowerPoint. In this essay he attacked slideware in general and PowerPoint in particular.

I read his essay with interest at the time, but found it hard to agree with many of his claims. I didn’t have a blog then to respond; now I do…

Examining Tufte’s Argument

Louis Gerstner example

Tufte starts with a quote from Louis Gerstner (Who Says Elephants Can’t Dance? Inside IBM’s Historic Turnaround). Gerstner describes reaching over and turning off an overhead projector in the middle of someone else’s presentation. He explained “Let’s just talk about your business.” He was surprised by the size of the impact it had on his staff members.

I, on the other hand, can fully understand such an act from a new company president causing consternation across the organisation. It sounds awfully rude!

Tufte claims that this action indicated that “there might be better ways to do serious analysis than reading aloud from projected lists”. That is a strawman. Of course there are better ways.

One way would be to do a full analysis and then present a summary, being careful not to just repeat what was on the slides but use it merely as a basis for the discussion. There was nothing in Gerstner’s story to suggest that wasn’t exactly the case.

However, a free-form discussion (“just talk”) without any reference points or visual aids is a particularly bad way to do a serious analysis. It leaves a serious risk that important facts or facets of the analysis will be forgotten.

Finally, this act changed the dynamics of the meeting – from a new staff member wanting to learn the local culture and the way things were done from the tribal elders, to a stranger who wanted to have equal input and make changes.

No wonder the staff members were concerned!

I don’t think Tufte has successfully made his point (which is presumably that PowerPoint is overly entrenched in the organisation).

Presenter-Oriented

Tufte suggests that PowerPoint is “entirely presenter-oriented and not content-oriented, not audience-oriented“. While I am not sure I understand the definitions of these terms, he goes on to argue that the “fans of PowerPoint are presenters, rarely audience members.”

This seems a bit feeble. PowerPoint is intended to help presenters do a good job of presenting to audience members. Audience members benefit indirectly, not directly. The audience don’t need to know what application is being used by the presenter.

How does it benefit the presenter? Tufte answers this himself with quotes of PowerPoint marketing claims.

Alliteration Ain’t Argumentation

I warned about the overuse of “distractions”, but explained that they serve a useful purpose… in moderation.

Tufte uses emotive terms like “chartjunk” and “PowerPointPhluff” (with meaningless illustrations) to dismiss them outright.

Low Resolution Leads to Brevity

Tufte complains that PowerPoint slides are low resolution. As you know, I absolutely agree – this is a serious limitation.

He starts with a strawman – a slide with only enough room for four words (“Correlation is not causation”). He then argues that this is a “seriously incomplete statement”, and suggests longer statements with more conditional clauses.

Even if we accept that the slide only has room for four words, perhaps such a bold, yet incomplete, statement still serves a purpose.

The bold statement can be softened through the words of the presenter. The bold statement serves to get the audiences attention. The bold statement ensures that the presenter and the reader remain on track, and don’t wander off topic. The bold statement may well be complete enough, if the presenter is arguing that causality is (incorrectly) being recklessly assumed in an example being discussed.

In short, the slides are not the presentation. The slides are the summary of the presentation.

Low Resolution Leads to Smaller Tables

Tufte then demonstrates that typical PowerPoint slides tend to have smaller data graphics (fewer entries per table) than a wide range of scientific journals and newspapers. To which I reply, “Thank goodness!”

If we consider the size of the tables involved in reporting stock prices in a newspaper and try to imagine it being presented on overhead slides, we can see that it would be unreadable.

However, such tables are designed to assist a large number of people with similar, but different, goals – each one is trying to extract a small amount of data from a large pool. They want to know how a handful of stocks are doing, not understand the entire pool.

If we want to understand the entire pool in the few minutes we have assigned to the problem within a presentation setting, we need to aggregate and summarise the data – e.g. a table of financial indicators.

The idea that PowerPoint slides do not have the same information as a reference document demonstrates appropriate use of the technology. If you want to handout complex reference material, do so in addition to the slides.

Bullets Points are not Silver

Tufte attacks the over-use of bullet points in PowerPoint presentations. He references another paper (reference) where they argue that bullet lists:

• are typically too generic
• leave critical relationships unspecified, and
• leave critical assumptions about how the business works unstated.

I think these arguments are quite valid and are actually on target. As I explained before, PowerPoint makes bullet points incredibly easy, but hasn’t had nearly the same effect on other layout styles.

However, Tufte ignores the key reason why this technique is chosen: saving limited time.

When that is taken into account, the decision to over-use bullet points may well be quite rational.

Space Shuttle Columbia Report

In a previous essay (Visual and Statistical Thinking: Displays of Evidence for Making Decisions), Tufte was critical of NASA’s analysis of the risk for launch of the space shuttle Challenger.

This time, he addresses NASA’s analysis of the tile damage caused by the debris that hit space shuttle Columbia. He explains that Boeing Corporation produced three quick reports and then lambastes them for producing reports that appearing to be quickly-produced!

Each of his criticisms is correct.

• Yes, the slides start with an executive summary that suggests that the risk is low, while the evidence didn’t support that.
• Yes, the language on the slide was clumsy to the point of misleading, and (horror of horrors!) there were typographical orphans.
• Yes, they abbreviated “cubic inches” in three different ways, and none of them the one that Tufte would prefer. (Nor were any the way I would prefer: cm³!)

Together, these slides failed to alert the management to the seriousness of the damage risk to the shuttle. The high-level NASA officials accepted the executive summary, and they made did not further analysis of the threat.

To Tufte, this suggests that the culture of PowerPoint is (at least partly) responsible for the resultant disaster. The PowerPoint slide failed to convey the engineering and modelling details to enable the high-level officials to perform the analysis.

To me, this suggests a completely different problem. The slides, despite showing signs of being cobbled together quickly in an emergency, achieved what I think was the real aim of the engineers: To convey the results of an analysis, not merely the ingredients; to persuade the high-level officials that a safe return of the shuttle was likely.

The analysis had already been completed. It appears to have drawn the wrong conclusion. Further examination may reveal that the analysis was flawed and tragically rushed, but that is where the focus should be.

The presentation, for all its clumsiness and sub-optimal features, was successful in conveying the result of the analysis to the high-level officials.

Tufte’s criticism of the language and format of the slides is further suggestive that the slides were rushed. However, in the exigent circumstances, it was seems quite rational not to spend much time on the niceties of copy-editing.

Genetic Fallacy

Perhaps the most ridiculous claim throughout Tufte’s essay is his attack on PowerPoint based on a confused metaphor with the companies who produced it.

This history [of PowerPoint being produced by an unnamed software house, and being acquired by Microsoft] is revealing, for the metaphor behind the PP cognitive style is the software corporation itself. That is, a big bureaucracy engaged in computer programming (deeply hierarchical, nested, highly structured, relentlessly sequential, one-short-line-at-a-time) and in marketing (fast pace, misdirection, advocacy not analysis, slogan thinking, branding, exaggerated claims, marketplace ethics). To describe a software house is to describe the PowerPoint cognitive style.

What rot! You may equally argue that most commercial software will stop working during the Christmas holidays or that the developers of “Tomb Raider” have breasts of an anatomically-unlikely size.

There are, indeed, some trends that you can elicit about software architectures based on team size and corporate culture, but I don’t think Tufte makes any credible case here. It may just be an analogy, but not a valuable one.

Cancer Rates

Tufte introduces a large table (with 196 entries) describing survival rates (and standard errors) for 24 cancer sites.

The raw data that Tufte provides is both very versatile and very dense. It would generally take a reader a large mental effort to process it into a form useful to them. In the context of a presentation, they would not have the opportunity, so it becomes necessary to summarise the data visually.

Tufte then puts it through the default graph generators in PowerPoint and mocks the result, suggesting it illustrates “statistical stupidity” and that “your audience will quickly and correctly conclude that you don’t know much about data and evidence.”

Damn right, but so what? If someone is foolish enough to perform these steps, they are being “statistically stupid”, and an audience is right to draw that conclusion.

Based on my experience, however, Tufte is setting up another strawman. While I have seen my fair share of meaningless charts, I don’t think PowerPoint has been the generator of many of them. PowerPoint graphics tend to be used only for the simplest of charts. For more complex charts – and the worst examples of meaningless charts – its near-neighbour, Microsoft Excel, tends to be used.

Tufte eventually presents an interesting and attractive solution to presenting the data. Sadly, it fails to meet many of the requirements.

Tufte fails to answer the question “What information does the reader want from this chart?”

If the reader is a cancer patient wanting to know their prognosis, the data should probably be presented in table form, sorted by cancer site.

If the reader is a researcher in epidemiology, they might want the data sorted by the standard error – the bigger the error, the more useful and promising a long-term study might be.

If the reader is an actuary, they might want the data in raw format for inclusion in their metrics.

If the reader is an accountant planning superannuation for a prostate cancer patient, they may just want to know the 20-year survival rate.

Tufte’s suggested format is an interesting hybrid between a table and a set of graphs. It emphasizes the cancer sites that have the greatest difference between the 5-year and 20-year survival rates. It does very poorly at prioritising between the sites by their 20-year survival rates.

Where it suffers the most is that, because of its novelty, it takes an unusually long time to recognise its structure; a useful tool for experts, perhaps, but not for occasional readers.

Proof There Exists a Bad Example

Tufte finds an example of a PowerPoint design template produced by a university facility. It includes guidelines on how to design slides.

The guidelines are, as Tufte suggests, rather witless. They over-emphasize the need to keep slides simple.

I suspect the guidelines were also probably written by people more used to a presenting to the lay public than to technical/scientific audiences. Hence they dumbed-down the text further than is desirable in the actual target setting: a medical school.

There was probably also some over-statement in an attempt to remedy a common presentation problem – the desire to cram too many words per slide, until it is unreadable.

Tufte takes a few unnecessary pot-shots: I think he is overly concerned with typographical orphans, he picks up on poorly-chosen initial caps, and he is bewildered by some ironic humour.

The template neglects to include the names of the presenters on the title slide. Unfortunate, but not, in my experience, typical of corporate design templates.

Ultimately, Tufte proves one thing. There exists at least one poor example of a PowerPoint design template. I could have proven that just by citing Sturgeon’s Revelation.

Data Density

Tufte spends most of three pages re-stating some obvious points that he had made before: there isn’t much room on a PowerPoint slide for a large amount of data; you are better off handing out printed copies of large tables.

It is worth noting, that handouts do require an extra level of effort: knowing how many attendees there will be, photocopying sufficient numbers, and filing the results. That may help explain why they are under-utilised.

Along the way, he slips in a valid criticism: that “lazy and ridiculous” people are using PowerPoint slides to replace rather than enhance the detailed report. I agree that taking analysis shortcuts this way is a risky approach.

Tufte also includes a brief debunking of the idea that the ideal number of bullet points is the classic “seven, plus or minus two”. He explains that the 1956 psychology study that is referenced to support this claim has been taken out of context here. Thank you, Tufte! I learned something here, and feel suitable chastened for relying on this factoid in the past.

Estimating the effects

Many corporate processes that we follow and tools that we use have a common characteristic: they greatly buoy up the poorest performers at a cost over extra, unnecessary, overhead for the best performers.

We adopt such tools and processes because being able to predictably produce satisfactory output is often more useful than unpredictably producing output with a large variance in quality.

Tufte agrees that PowerPoint has this characteristic:

In day-to-day practice, PowerPoint templates may improve 10% or 20% of all presentations by organizing inept, extremely disorganized speakers, at a cost of detectable intellectual damage to 80%.

Perhaps if Tufte’s unsubstantiated figures were correct, PowerPoint would not be a worthwhile tool. However, I think that he, as a top-performer in presentations, under-estimates the number of presenters who are inept, and under-estimates the number of speakers who are “ept”, but have insufficient available time to appear anything but disorganized. If PowerPoint was to improve 60% of presentations, and to only damage 5%, it might be seen as a valuable asset.

I encourage the best speakers to eschew the use of slideware if they really can perform better without it. However, for the typical untrained and unskilled presenter in a business communications format, PowerPoint can help you make a poor presentation into a satisfactory one.

Winding up the Tour

Tufte and I have both been bored and frustrated by poor presentations. We both lament shortcuts in the analyses performed and minimal level of understanding that managers seem to require to make serious decisions.

The difference between us is that Tufte lashes out at Microsoft, where I share blame amongst a triumvirate: lack of time, lack of skill and lack of screen-resolution.

Clarifications

Microsoft Apologist?

As I mentioned before, I am afraid of sounding like a Microsoft apologist – that’s not what I am.

I have my own litany of complaints about PowerPoint. After just a couple of minutes thought, I came up with:

• the too-frequent crashes
• the frequent corruption of bullet levels, so that the bullet-styles don’t match
• the (related) inability to “reveal codes”
• the lack of “newspaper-style columns” functionality
• the bizarre “toggling” interaction between bullet points and indent levels
• the horrid auto-scaling grids that make the data look awful, every single time

Each of these is annoying – perhaps even worth a snarky blog article or two – but none of them can be seriously blamed for causing space shuttles to drop from the sky.

Hater of Management?

I want to be quite clear that the comments on managers are not a personal attack on managers I work with today. My comments are based on a career of being on both sides of the presentations. I first got fired up by Tufte’s comments over two years ago – my managers have changed since then.

I also don’t want my comments about managers to be taken as a general tirade against management. I think that I have shown that often the decisions being made by busy managers are rational given time constraints. I also plead guilty to having committed just about every one of these misdemeanours myself.

Late on the Scene

Tufte published his essay almost three years ago. I am not the first to respond. I could have done more research early on to find out what other people have said about the essay, but I didn’t. I wanted to think through the issues myself, and work out why Tufte’s claims had got to me fired up, rather just nod at what someone else said, and forget it.

There are other critics of Tufte who beat me to it. I searched for a few after finishing a draft of this essay. Some examples include:

• Don Norman’s In Defense of PowerPoint
• Loisel and Galer’s Uses of PowerPoint in the 314L Classroom
• Tad Simons’ Does PowerPoint make you stupid?

Reading the latter, I learnt that:

For designers who agree with Tufte on many things, but disagree vehemently with his opinions on PowerPoint, parsing the great man’s argument has become something of a parlor sport.

Whoops! I had no idea I was partaking in such a popular endeavour. Sorry!

Further, I have not done any searching to find out whether Tufte has responded to this criticism in the meantime. Someone seeking a balanced understanding would do well to get a copy of Tufte’s original essay and check for his responses to criticisms such as mine.