Should I write an article or have a presentation?

It all started one day when my boss called me to ask me a few questions about one of the ongoing projects. When I questioned his decision on having a meeting when a few sentences long email would do just fine, I was told that a meeting takes less time. I’ve met plenty of coworkers who likewise claimed how meetings resolved problems faster, but is that really so?

If we take a look at this problem strictly from the data transfer point of view, then the written text will always win over any other form of communication, because reading is unfortunately still the most efficient way of transferring knowledge that we know of today. The reason why I say “unfortunately” is due to its relatively low bandwidth.

An average reader is only able to read around 230 words per minute (WPM). Considering that an average word in an English article is 5-6 characters long and each character is 1 byte in size, we can quickly calculate that an average reader ingests information at a rate of roughly 1380 bytes per minute. If the internet service providers throttled your internet connection to this number, since there is no way your brain could process more information, I think most of you would be pretty pissed off as modern websites would take forever to open.

But how does reading compare to spoken lectures or meetings? Surely one can speak faster than they can read? It turns out that is not the case, since an average speaker is speaking at the rate of around 150 words per minute which evaluates to a transfer of roughly 900 bytes per minute. Most speeches are also riddled with expletives, which further decreases the amount of useful information that you receive in a minute.

If reading makes for a faster way of transferring knowledge, but it takes more time for the writer to write the content, at what point should we prefer to opt for writing an article over delivering a public lecture?

We can define a simple equation and get to the bottom of this problem right away:

content = preparation + writing_time + readers * reading_time
content = preparation + presenting_time + listeners * listening_time

We can cross out the preparation time, since we estimate that it takes roughly the same amount of time for preparing a good lecture as it does for writing a thorough article. Consequently our equation shrinks to the following:

content = writing_time + readers * reading_time
content = presenting_time + listeners * listening_time

The math nerds claim that from two related equations we can calculate at most two unknown variables, but in our equations we have many more. Luckily we know, however, that presenting_time is the same as listening_time and we know that presenting_time, listening_time, writing_time, reading_time are all related to the content variable. To make this concept a bit clearer, we are going to use a specific example for our content: a 4000 words long article 1.

The following table contains the numbers which are going to be used in our calculation:

Action Speed
Average typing speed 40 WPM
Average reading speed 230 WPM
Average talking speed 150 WPM
Listening speed 150 WPM
# Writing time calculation
4000 words = writing_time * 40 WPM
writing_time = 4000 / 40 
writing_time = 100 minutes

reading_time = content / 230 WPM
reading_time = 4000 words / 230 WPM
reading_time = ~17 minutes


# Presenting time calculation
4000 words = presenting_time * 150 WPM
presenting_time = 4000 / 150
presenting_time = ~27 minutes

listening_time = presenting_time
listening_time = ~27 minutes

Now that we have some actual numbers, we can combine the previous equations and calculate the necessary number of people that our content has to reach before written article becomes a more effective way of transferring knowledge than a spoken lecture:

writing_time + readers * reading_time = presenting_time + listeners * listening_time
100 minutes + readers * 17 minutes = 27 minutes + listeners * 27 minutes

# we want to know at which number of receivers these equations
# are equal, thus we replace 'readers' and 'listeners' with 'x'
100 minutes + x * 17 minutes = 27 minutes * x * 27 minutes
100 minutes - 27 minutes = 27 minutes * x - 17 minutes * x
73 minutes = 10 minutes * x
x = ~8

From the calculation above we can deduce that reading becomes a more efficient way of transferring knowledge when you have at least 8 readers. This rather simple calculation therefore reveals an important insight: if you want to educate a large number of people on some topic, you should write an article or book since written text just scales better.

Since you are reading this article on the web, we can leverage the power of modern technology and make this calculation right here in your browser, so you can play with these numbers on your own:

Making this calculation, however, raises a few questions. If writing is a much more efficient way of transferring knowledge, how come there are so many people who prefer to be the talkers over being the writers? I think there are a couple of reasons:

Preparation

Meeting organizers usually don’t prepare for meetings in advance or they spend very little time on the preparation, which leads to them saving a considerable amount of time; judging from the calculations above, it’s clear that for small groups the preparation time will dominate the total time spent on the transfer knowledge process.

What is concerning, however, is that the very same organizer will like to skip on preparation even when it comes to a meeting with a larger group of people (e.g., department wide meeting). Because most people optimize locally without looking at the big picture, they will naturally prefer to save a few minutes of their preparation time over saving many more minutes of their listeners' time by delivering a terser/better prepared lecture.

Typing

A lot of office workers don’t know how to type well, therefore having to write a few pages of text is going to take them a lot of time. Some like to argue, however, how much of their time is spent on thinking about the topic rather than on typing that thinking down.

While I understand that people work in a number of different ways, and what works for me does not work for somebody else, I figured out that typing is certainly a bottleneck in my case. When I am writing any kind of long form text such as an email, comments in the source code or some kind of a long form document I am most often restricted by my typing ability. There are two reasons why I feel that way:

  • Most of these typing tasks are not very brain intensive. Let’s be honest here: most of us are not writing scientific articles that demand a high density of information due to the predefined number of pages into which an article has to fit in. Most of the regular job writing is fairly repetitive and once you get a hang of it you can write mostly on autopilot.

  • You will rarely get anything right the first time regardless of what you are working on, be it software, documentation or what have you. In order to produce a high quality thing you will have to iterate on it. During the iteration process your brain will usually produce flashes of insights and ways to rephrase existing sentences which will sound much better than whatever you have come up with when writing your first draft. If you can keep up with your brain when you are iterating on your text, you will be able to go through more iterations and consequently produce higher quality content.

To be more concrete about the mentioned typing bottleneck problem, let’s have a real life example. My usual typing speed is around 100 WPM. In comparison to a hunt and pecker (~30 WPM), I type 3.3 times faster. For typing a small email, such as a weekly progress report that contains 200 words, I am done writing it in about 2 minutes. The hunt and pecker, however, will have to spend close to 7 minutes to accomplish the same task.

Throughout your day this writing time accumulates which is exactly why slow typists tend to prefer having a conversation. If we assume that a talking speed is around ~150 WPM, they would be finished with their progress report in about 1.5 minutes which is a considerable reduction. For this reason poor typists rarely produce descriptive emails, detailed bug reports or any kind of long form documentation. Their iteration process is slower, therefore their emails are short, bug reports are lacking details and the documentation is often skipped. After all, there are only so many hours in the day.

When is typing a waste of time?

Before we declare a victory of writing over public speaking, I think we should also note down occasions when spending time on writing is more trouble than it’s worth.

Small number of participants

Some meetings don’t have a lot of participants (such as 1 on 1 meetings), or they end up as unrehearsed brainstorming sessions for which it’s hard to assign a time related value. A long meeting with a few key insights thrown in the middle, might be worth more than saving a few minutes of participants time 2.

At what point is it worth for a hunt and pecker to write instead of having a conversation? If you enter the numbers in the calculator above, you will learn that they need an audience of at least 12 people before they will find typing a worthy activity.

Discussions

Written material is usually a one way street: I, the writer, am the authority and you shall all listen to me with no way to ask further questions or complain about incorrect information. On the other hand, lectures let you ask questions about confusing parts of the presentation. Even if the lecturer doesn’t know the answer, sometimes someone from the audience might know it. Such impromptu collaboration can’t happen with a written text.

That said, some people will need a few days to process the content before they will be able to ask a meaningful question. Nevertheless, in a fast moving lecture you simply don’t have that kind of time available.

How many people should read this post?

According to my little time tracking experiment, I noticed that 1 hour of writing produces roughly 1 minute of reading (all revisions and illustrations included). Given that a minute of reading time takes 230 words, my average article writing speed sits at roughly 4 words per minute.

Entering this number in the calculator above will tell you that each of my posts should reach at least 105 people before I am able to say that writing an article was worth it. Considering how often my posts caused an internet wide food fight (see blog traffic article), I can say that writing these kinds of articles was a net positive activity, if and only if we can say that my drivel brings value to the one reading it.

If a picture is worth a thousand of words, shouldn’t a comic book be an even better choice?

Well, yes in theory at least. The only problem is that the time spent for producing such a book would sky rocket, since drawing is a very time consuming process. How would a dense math book look like in a comic book format is still something to ponder about. After all, if comic books would simplify the knowledge transfer process so much, then why are the academic papers still written as a regular boring texts?

Part of the reason is due to tradition, and part of the reason may be simply because we haven’t figured out a better way.

Appendix A: Typing speeds

Typing Experience Speed Percentile
Hunt and pecker 27-37 WPM 10-30
Average office worker 40 WPM 37.4
Above average office worker 50 WPM 58
Productive typist 60 WPM 74
Experienced typist 80 WPM 93
Top 2% 100+ WPM 98.4

Percentiles were taken from this typeracer analysis (2017). If the original post is no longer available, you can check out the archived screenshot.

Notes


  1. The length of the article is irrelevant in this calculation. A careful reader might even realize that we don’t have to calculate anything to get to the same conclusion. If we assume that our content will be consumed by a large number of people, then surely the time spent on consuming the content will be much larger in comparison to the time that the author had spent to produce the content. For this reason we can cross out the time of production on both sides of the equation, and we are left with:

    readers * reading_time = listeners * listening_time
    

    Since we have already determined that reading is faster than listening, we already know that written content is a more efficient way of transferring knowledge. ↩︎

  2. From my experience, however, meetings without a strict agenda usually ended up being a waste of time. ↩︎